Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

IA-GCN: Interactive Graph Convolutional Network for Recommendation

Yinan Zhang, Pei Wang, Congcong Liu, Xiwei Zhao, Hao Qi, Jie He, Junsheng Jin, Changping Peng, Zhangang Lin, Jingping Shao

TL;DR

IA-GCN tackles the suboptimal late fusion in GCN-based recommendation by introducing bilateral interactive guidance between the user and item trees, conditioning neighborhood aggregation on the target counterpart to preserve target-specific information throughout propagation. The method preserves a light, transformation-free aggregation similar to LightGCN but with interactive weights computed from 0th-order embeddings to emphasize neighbors related to the target, and combines layer features with learnable coefficients. Empirical results across Gowalla, Yelp2018, and Amazon-Book show IA-GCN outperforms LightGCN and is competitive with or superior to state-of-the-art baselines, with notable gains on sparse data like Amazon-Book. The work demonstrates that interactive, cross-tree guidance can significantly enhance expressiveness and interpretability in CF, offering a plug-in module that can improve existing GCN-based recommendation systems.

Abstract

Recently, Graph Convolutional Network (GCN) has become a novel state-of-art for Collaborative Filtering (CF) based Recommender Systems (RS). It is a common practice to learn informative user and item representations by performing embedding propagation on a user-item bipartite graph, and then provide the users with personalized item suggestions based on the representations. Despite effectiveness, existing algorithms neglect precious interactive features between user-item pairs in the embedding process. When predicting a user's preference for different items, they still aggregate the user tree in the same way, without emphasizing target-related information in the user neighborhood. Such a uniform aggregation scheme easily leads to suboptimal user and item representations, limiting the model expressiveness to some extent. In this work, we address this problem by building bilateral interactive guidance between each user-item pair and proposing a new model named IA-GCN (short for InterActive GCN). Specifically, when learning the user representation from its neighborhood, we assign higher attention weights to those neighbors similar to the target item. Correspondingly, when learning the item representation, we pay more attention to those neighbors resembling the target user. This leads to interactive and interpretable features, effectively distilling target-specific information through each graph convolutional operation. Our model is built on top of LightGCN, a state-of-the-art GCN model for CF, and can be combined with various GCN-based CF architectures in an end-to-end fashion. Extensive experiments on three benchmark datasets demonstrate the effectiveness and robustness of IA-GCN.

IA-GCN: Interactive Graph Convolutional Network for Recommendation

TL;DR

IA-GCN tackles the suboptimal late fusion in GCN-based recommendation by introducing bilateral interactive guidance between the user and item trees, conditioning neighborhood aggregation on the target counterpart to preserve target-specific information throughout propagation. The method preserves a light, transformation-free aggregation similar to LightGCN but with interactive weights computed from 0th-order embeddings to emphasize neighbors related to the target, and combines layer features with learnable coefficients. Empirical results across Gowalla, Yelp2018, and Amazon-Book show IA-GCN outperforms LightGCN and is competitive with or superior to state-of-the-art baselines, with notable gains on sparse data like Amazon-Book. The work demonstrates that interactive, cross-tree guidance can significantly enhance expressiveness and interpretability in CF, offering a plug-in module that can improve existing GCN-based recommendation systems.

Abstract

Recently, Graph Convolutional Network (GCN) has become a novel state-of-art for Collaborative Filtering (CF) based Recommender Systems (RS). It is a common practice to learn informative user and item representations by performing embedding propagation on a user-item bipartite graph, and then provide the users with personalized item suggestions based on the representations. Despite effectiveness, existing algorithms neglect precious interactive features between user-item pairs in the embedding process. When predicting a user's preference for different items, they still aggregate the user tree in the same way, without emphasizing target-related information in the user neighborhood. Such a uniform aggregation scheme easily leads to suboptimal user and item representations, limiting the model expressiveness to some extent. In this work, we address this problem by building bilateral interactive guidance between each user-item pair and proposing a new model named IA-GCN (short for InterActive GCN). Specifically, when learning the user representation from its neighborhood, we assign higher attention weights to those neighbors similar to the target item. Correspondingly, when learning the item representation, we pay more attention to those neighbors resembling the target user. This leads to interactive and interpretable features, effectively distilling target-specific information through each graph convolutional operation. Our model is built on top of LightGCN, a state-of-the-art GCN model for CF, and can be combined with various GCN-based CF architectures in an end-to-end fashion. Extensive experiments on three benchmark datasets demonstrate the effectiveness and robustness of IA-GCN.
Paper Structure (29 sections, 11 equations, 4 figures, 7 tables)

This paper contains 29 sections, 11 equations, 4 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Collaborative Filtering
  4. GCN for Recommendation
  5. Attention Mechanism
  6. InterActive Graph Convolutional Neural Network
  7. Preliminaries
  8. Graph Convolution Framework
  9. Interactive GCN for Recommendation
  10. What is "Interactive"?
  11. Interactive Guide
  12. Interactive Convolution
  13. Layer Combination and Model Prediction
  14. Model Analysis
  15. Relation with LightGCN
  16. ...and 14 more sections

Figures (4)

  • Figure 1: The two-tree structure in GCN-based recommendation framework
  • Figure 2: The proposed InterActive GCN structure.
  • Figure 4: Training and Testing evaluation metrics curves of LightGCN and IA-GCN. All methods are evaluated by Recall@20 ranging from 100 to 1000 epochs on Yelp2018. For fairness comparison, we conducted experiments on Gowalla and Yelp2018 under the learning rate $5.5e^{-5}$ and $6.5e^{-5}$ respectively.
  • Figure 5: Results of IA-GCN and the variant that uses weighted layer combination at different layers on Gowalla, Yelp2018, and Amazon-Book